Process times of severely injured patients in the emergency room are associated with patient volume: a registry-based analysis

It is a commonplace that with more practice and experience performance will improve. This holds also true in medicine. Surgeons know that there is a learning curve if they start a new operation technique. For the care of severely injured trauma patients there are, however, conflicting results. While some studies did not observe such a relationship [1‐4], others reported a positive correlation of patient volume per trauma center with mortality and other outcome measures [5‐8].

Up to now, such analyses investigating the effect of patient volume in trauma care mainly came from North America. Particularly for trauma systems with smaller sized trauma centers, as in Germany, such investigations were missing. In Germany it has been shown, however, that mortality rates differ between large and small hospitals but the standardized mortality ratio (observed vs. predicted mortality) is similar [9].

Nevertheless, this assumption led to the inclusion of an Advanced Trauma Life Support (ATLS) training for the emergency room team members into the list of criteria required to become a certified trauma center in Germany and other countries [10]. But such training sessions could only be a starting point. Knowledge and skills need permanent training and practice. However, not all hospitals could provide such a condition because of too few trauma patients.

There are nearly 700 German hospitals certified as a trauma centers. Three levels were defined: local (level 3), regional (level 2), and supra-regional (level 1) trauma centers, based on the availability of resources and, for level 1 and 2 centers, also a minimum number of severe trauma cases per year. These criteria are checked every 3 years at an on-site audit visit.

Annually, there are about 18,000 severe trauma patients in Germany, defined as Injury Severity Score (ISS) ≥ 16 [11]. About 33,000 trauma patients with maximum Abbreviated Injury Scale (AIS) severity ≥ 3, or with need for intensive care, are documented per year in the TraumaRegister DGU^® (TR-DGU), including the above mentioned severe trauma cases. These cases were not evenly distributed among the 700 trauma centers. The majority of cases were treated in about 100 level 1 trauma centers. But a considerable number of trauma cases were transported to a small level 3 center. Such centers often see less than ten cases per year, which means, less than one patient per month.

The present analysis does not aim to answer the question whether the outcome of severely injured trauma patient treated in a small hospital is worse than in a large trauma center. It is rather the aim of this analysis to describe the effect of low, medium, and high patient volume on the initial process times in the emergency room. Although everyone assumes that such a training effect exists, there is only scarce knowledge about the amount of this effect, or the critical number of cases per year where this effect becomes important. The results could thus be an empirical contribution to the discussion about the appropriate target hospital for severely injured patients.

Hospitals were grouped into five subgroups according to their annual number of severely injured patients admitted primarily to the emergency room of each hospital. Most patients were treated in the 78 large hospitals with more than 100 cases per year (n = 58,474; 45.3%). There were 203 hospitals (30%) with less than 10 cases/year. In these 203 ‘small’ hospitals 3956 patients (3.1%) were treated. Table 1 shows the demographic and injury data of the five subgroups of hospitals.

Patient characteristics are presented in Table 1. Patients admitted to high volume hospitals tended to be younger, were more severely injured, had more head injuries, and their hospital mortality was higher. Every fifth patient (19%) admitted to a small hospital with less than ten cases per year was transferred out to another hospital within a few hours.

There were also differences in the frequency of using diagnostic procedures. While abdominal ultrasound (Focused Assessment with Sonography for Trauma, FAST) and chest X-ray showed approximately similar prevalence rates, there was a clear shift in using computed tomography (CT). The use of whole-body CT increased from 59% in the smallest hospitals to 83% in the large ones. The average times since admission needed to start these investigations are presented in Fig. 1. If time to first CT was limited to cases with GCS < 14 only (38% of all cases) average time intervals did not differ from the overall times (difference within a range of ± 1 min per hospital category).

If smallest and largest hospitals were compared, time to whole-body CT was reduced by about one third (9 min. faster in large hospitals), time to FAST was approximately halved (4 min. faster), and time to first X-ray was even more than halved (17 min. faster).

For a set of eight surgical emergency procedures, time intervals were documented separately in TR-DGU. Those procedures were more frequently performed in hospitals with a higher annual volume, and thus higher injury severity. The mean time to the first procedure increased with volume: in small hospitals it took 60 min on average (SD 33), while it took 74 min (SD 27) in hospitals with the highest volume. However, this trend is not identical for all emergency interventions. Figure 2 shows that this delay is mainly caused by external stabilization of pelvis and extremities.

The median time to start blood transfusion was 35 min in large hospitals (100 + cases/year) and 47 min in small hospitals.

Table 2 describes the total time in the emergency room (ER). Overall, there is a considerable and consistent decrease in length of stay in the ER with increasing volume (Table 2 and Fig. 3).

787 patients (1.6%) died in the ER; their median time to death was 33 min, and this time did not differ across the hospital subgroups. Another group of patients were directly transferred from the ER to another hospital. Their fraction decreased from 22% in hospitals with lowest volume to less than 1% in hospitals with highest volume. On average, they stayed about 2 h in the ER before transfer, with only minor variation across the hospitals. The largest group of patients was discharged from ER to ICU. Their length of stay in the ER only slightly decreased with increasing volume (on average 10 min faster in high volume hospitals). The largest time savings were observed in patients transferred to the operation room for surgery. Not only the percentage of cases discharged to the OR was highest in large hospitals (29%), but also the time in the ER was shortest (mean time 65 min, Table 2). This value also had the lowest SD indicating that there were less outliers with a long delay.

The median prehospital time from the accident to hospital admission was 9–12 min longer for large hospitals treating 40 cases/year or more (Fig. 4). There is a net time saving of about 10 min (from the accident to the end of ED treatment) for large hospitals if the time in the ED is added.

There is an ongoing debate whether severely injured patients should be brought to the closest hospital nearby to save time, or to directly admit such a patient to the next ‘appropriate’ trauma center with adequate resources, even with bypassing the closest hospital. Those who promote the second option argue that the additional time required for transportation will easily be compensated by a more efficient acute care in specialized centers. Some people even promote to establish ‘mega centers’ with several hundreds of severe trauma cases (ISS 16+) per year. However, too long transportation times may also result in a case selection bias excluding instable trauma victims who quickly died.

The present analysis is based on a large nation-wide trauma registration in Germany. We do not have such ‘mega centers’ here and could therefore not contribute to that discussion. However, we have a large number of hospitals (about 700) participating in the treatment of severely injured patients. And most of these hospitals only see a limited number of patients per year. This would allow comparing process times in the acute phase even if the patient volume is low.

We classified hospitals by their annual number of severely injured patients (based on a 5 years average) which is more specific than using the certified level of care. Since our documentation also includes the pre-hospital setting (times, physiology, and interventions), we were able to quantify the additional time needed to bring a patient to a high volume hospital with appropriate resources.

As expected, there is a relation between patient volume and process times. This relation is most prominent for time to first CT, first X-ray, and overall time in the ER, especially for patients transferred to the operation room. But we were not able to define an appropriate cut-off point regarding patients per year. The considered times still decreased up to the highest volume subgroup of hospitals. For very large hospitals we did not have enough data; there were only nine hospitals with > 200 patients/year. This is a clear argument that frequent training will reduce process times. Interestingly, the average time until CT-scanning was 20 min in the highest volume centers and 29 min in the lowest volume centers in our study, while the median time until CT-scanning was reported to be 82 min in a Dutch very high-volume trauma center [13]. Therefore, international comparisons with results from different health care systems have to be interpreted with caution.

Interestingly, not all emergency interventions showed this trend. The time to some acute interventions did not vary across the hospitals (e.g., laparotomy with about 70 min). Time to other interventions, like external fixation, even took longer time in large hospitals. This might be caused by prior treatment of other life-threatening injuries, or a successful stabilization and resuscitation, which reduces the urgency. Other interventions like a thoracotomy showed a tendency towards a faster performance. While it is not clear whether time savings until diagnostic laparotomy lead to improved outcome [14] a shorter time until emergency thoracotomy correlates with a reduction in mortality [7].

On the other hand, there is a discussion about the additional time needed to bring a patient directly to a large trauma center. According to our results, this difference is rather small (about 10 min), compared to the time savings in the ER. The longer pre-hospital time could be explained by the higher injury severity requiring additional interventions at scene, and the use of helicopters for transportation [15]. This is mainly because the helicopter is usually not the primary vehicle but will be dispatched on request by the ground EMS team. The pure transportation time from scene to hospital is only 2 min longer in cases admitted to large hospitals. This delay is compensated already after the first few minutes in hospital.

It was not our aim to evaluate the relationship of patient volume and outcome in this paper. Such a relationship is often assumed, and it has also been shown by several authors in trauma care [8, 16‐19], while other authors could not prove such a relationship [3]. Previous analyses from TR-DGU in ISS 16+ patients (without transfers) suggested an adequate severity-adjusted outcome if more than 40 patients were treated annually [20]. In the present study, however, observed and expected mortality were nearly identical in all subgroups, if transfers were excluded. But we could not answer the question whether initial stabilization in a small hospital with subsequent transfer, or direct admission to a large trauma center is associated with a superior outcome. This is because our registry system does not allow matching treatment phases in different hospitals with sufficient precision. But if only process times were considered then longer travel times to a large trauma center were more than balanced after the initial shock room phase. This requires, however, a dense distribution of large trauma centers, as given in Germany.

Process times are frequently used as quality indicators (QI). A recently performed expert rating of 40 QI favored such process times at least for whole-body CT and several emergency interventions [21]. The subsequent empirical validation of these QI using TR-DGU data found a stronger association of event-related QI rather than process times [22]. In other words, it is more important that a certain intervention is performed, rather than it is performed 10 min earlier or later.

Our investigation bears all limitations of registry studies: there is no verification of correctness, and missing values were not queried. The exclusion of cases with missing or implausible time data may thus induce a bias. However, the online documentation software has a large number of plausibility checks implemented. The TR-DGU is well-known in Germany, and runs already for 25 years, and process times in the ER have been collected since the beginning. Amalgamation of treatment phases from different hospitals could not be performed due to a missing unique case identifier in Germany. This limits the outcome analysis of transferred cases.

Furthermore, the present results are valid for the German trauma system only which has some specific characteristics: nearly all trauma victims were seen by a physician at the scene who decides about the transport destination. Traditionally, the number of hospitals engaged in trauma care is high which leads to a large number of hospitals involved but with low patient volume per hospital. On the other hand, all participating hospitals were certified by the German Trauma Society and have shown to fulfill the requirements as laid down in the national Whitebook for the Medical Care of the Severely Injured [23]. Among other criteria they provide 24 h availability of CT, defined trauma teams, ATLS training for at least 50% of physicians in the trauma teams, etc. Therefore, a high level of expertise is present even in certified lower volume hospitals. Furthermore, population density is generally high in Germany so that a hospital could be reached within a few minutes, in most areas. An extended air rescue system with 89 helicopters further helps to reduce the pre-hospital time.

For severely injured patients, there is a clear and linear association of patient volume and process times in the emergency room, including several diagnostic procedures and the overall time in the ER. Up to 20 min could be saved per patient in large hospitals with more than 100 severe trauma cases per year. On the other hand, the increase in pre-hospital transportation time is less distinct.